Chemical Database Identifiers (PubChem, DrugBank, UniProt, InChIKey)
PubChem CIDs, DrugBank IDs, UniProt accessions and InChIKeys each name a different object, a structure, a drug entity or a precursor protein, and cross-checking them pins down which molecule a claim concerns.
A PubChem CID names one standardised structure, so a peptide and its acetate salt are different CIDs. A DrugBank ID, DB plus five digits, names a drug entity across its salt forms and records its status. A UniProt accession such as P01275 names a gene-encoded protein as translated, meaning the precursor, with mature hormones annotated within. An InChIKey is a 27-character hash of a complete structure, connectivity in the first 14 characters, stereochemistry and isotopes in the second block.
P01275 is human proglucagon; glucagon, GLP-1 and GLP-2 are all cut from that entry, so the accession alone does not say which. Semaglutide, with its Aib residue and C18 diacid, is encoded by no gene and has no accession; its identity lives in its CID, DrugBank DB13928 and its InChIKey. A diastereomer with one D-residue shares the first 14 InChIKey characters and differs only after the hyphen.
Identifiers are pointers to follow, not badges. Resolve each and check the structure it returns against the label wherever products diverge: sequence, amide or acid terminus, salt form, and a molecular weight matching the formula. Agreement among CAS number, CID, InChIKey and formula shows only that the description is internally consistent; only an analytical method says what is in the vial.
A synthetic analogue listed under the native hormone's UniProt accession is being passed off as the human peptide; the accession names a precursor it was never cut from. A DrugBank ID implies no approval; investigational compounds have IDs too. A CID that resolves to the free acid for a product described as the amide, or to the acetate salt beside a free-peptide molecular weight, is a 1 Da or 60 Da mismatch the identifiers exist to catch.